2 The faint I-band ID is confirmed.
  4 The northern of the two I-band candidates has been spectroscopically 
  4  confirmed as the host galaxy.
  7 This faint ID is confirmed in the K band and also spectroscopically.
 13 An ID for the host galaxy has been found on the SE component. As discussed 
 13  in Paper 1, it is likely that the NW radio components are not associated 
 13  with this and that we have a single source. In that case, those two NW 
 13  sources do not have sufficient flux to be included in the sample. The radio
 13  flux density given for this source, in Paper 1, is for the SE component 
 13  only.
 14 The faint I-band ID is confirmed.
 20 There is a potential K-band ID which is a red source associated with the 
 20  northern radio component. However, spectroscopy (see Paper 3) reveals that 
 20  emission lines are associated with the centre of the southern component, 
 20  indicating that the true host is located there. The appropriate magnitude 
 20  here is then the limiting magnitude at that point, which is 19.4.
 28 I-K~3. For a low-z galaxy, which must be from its K-band magnitude, this is 
 28  a very red colour which reinforces this candidate as the likely host galaxy
 31 The western optical source is a star. The eastern candidate has not been 
 31  spectroscopically confirmed but has I-K~2.7 which is not inconsistent with 
 31  an AGN host at its Kz estimated redshift.
 32 The faint I-band ID is confirmed.
 37 The faint I-band ID is confirmed in K and also spectroscopically.
 38 Whilst this candidate was not found to be likely according to the analysis 
 38  of Paper 1, subsequent spectroscopy shows that it is a quasar and is 
 38  therefore the correct identification for the radio source host galaxy.
 42 The faint I-band ID is not detected in this (shallow) K-band observation but
 42  has a single emission line (see Paper 3) and is therefore correctly 
 42  identified as the AGN host.
 43 The faint I-band ID is confirmed in K and also spectroscopically.
 47 The I-band ID is confirmed in K and also spectroscopically.
 57 Whilst no spectroscopic confirmation has been obtained, the red, I-K~3.7 
 57  colour, of this galaxy provides additional evidence that it is indeed the 
 57  host galaxy.
 58 There is a very tentative detection of a host galaxy for this source, 
 58  however, a nearby source dominates emission.
 60 The I-band ID is confirmed in K and has I-K~4.5. This red colour is 
 60  indicative of a 1<~z<~2 elliptical, consistent with an AGN host.
 62 The faint I-band ID is confirmed in K and has I-K~3, which is red for its 
 62  expected redshift.
 67 The I-band ID is confirmed in K and has I-K~2.3.
 69 There is a faint source on the I-band image. However, it was not identified 
 69  as a likely candidate. There is no detection in the K band so a K-band 
 69  limit is measured.
 70 This is a large extended radio source for which it was not possible to 
 70  easily associate a host galaxy based on the I-band imaging from EIS. K-band
 70  imaging allows candidates to be selected on the basis of colour. These are 
 70  presented in Table 3. Since CENSORS 70 is a fairly large radio source 
 70  (3arcmin) it is likely that it is at relatively low redshifts (at z>1 
 70  this angular size corresponds to ~1Mpc). Thus candidates B and E, which 
 70  have IK typical of elliptical galaxies at z<1, are, by this argument, the 
 70  most likely candidates. This issue is unlikely to be resolved without 
 70  spectroscopic follow-up of all these candidates.
 74 There were no likely candidates according to the analysis of Paper 1; 
 74  however, there are two optical sources visible in the I-band image which 
 74  might be associated with radio source. K-band imaging reveals that the more
 74  eastern of these two is a mildly red (I-K~2.9) source and is adopted as the
 74  candidate.
 81 The host galaxy is detected in K and has been spectroscopically confirmed.
 82 The identification came from the I band, but the spectrum shows this to be a
 82  star. There is no host galaxy candidate from the K band.
 84 As discussed in Paper 1 this radio source (EISD 103) was associated with 
 84  EISD 73 and 151. It is believed that EISD 73 and 151 were correlated noise 
 84  in the NVSS data thus leaving EISD 103. The higher-resolution maps then 
 84  revealed this to be two radio sources leaving the possibility that either 
 84  they were associated, and would therefore have sufficient combined flux to 
 84  be included in CENSORS, or that they were unassociated. The K-band imaging 
 84  reveals that there is no red galaxy between the two radio sources that 
 84  might indicate that they are associated and are radio lobes. There is a 
 84  candidate host galaxy associated with the western radio source. This is 
 84  taken to be the correct host galaxy for this source and it continues to be 
 84  included in the CENSORS sample on the assumption that the eastern source 
 84  is associated with it.
 86 The host galaxy is detected in K and, whilst not spectroscopically 
 86  confirmed, I-K=5.3 and this is typical of the old elliptical galaxies 
 86  which host AGN.
 89 There are two K-band candidates for this source. The eastern identification 
 89  emits an [O II] line and shows strong H and K absorption features 
 89  (see Paper 3).
 90 As discussed in Paper 1, CENSORS 90 and 103 (EISD 114 and 56, respectively) 
 90  may be associated (CENSORS 10/EISD 16 is nearby but is clearly associated 
 90  with a double radio source). Fig. A2 shows the I- and K-band images with 
 90  the radio contours. Component A is CENSORS 10. It was suggested that the 
 90  optical counterpart to component F was evidence that it was a single 
 90  source, too weak to be included in the CENSORS sample. This is consistent 
 90  with the counterpart being a red source, detected in the K band. B may 
 90  correspond to CENSORS 90 while component E corresponds to CENSORS 103; 
 90  component D has a counterpart and causes the NVSS map for CENSORS 103 to be
 90  extended to the NE. However, neither component B nor E has a counterpart in
 90  either the optical or near-IR, whereas component C does and lie between the
 90  two, raising the possibility that B, C and E are associated.
 94 The off-axis NE source is believed to be the host galaxy. The other 
 94  candidate is a star and the NE source has I-K>4.
 97 I-K~3.2 confirms this as the likely candidate.
103 See CENSORS 90.
104 There is no spectroscopic confirmation but this has I-K~5.3.
105 This red (I-K>3.3) host galaxy candidate is confirmed by extended Ly{alpha}
105  emission (Paper 3).
107 This galaxy has I-K>2.6, and has been shown spectroscopically to be at z~0.5
107  (Paper 3). The spectrum is of an old galaxy, and the colour is red for that
107  redshift, suggesting that this is indeed the host.
109 With I-K>2.9, this is a relatively red galaxy for its expected redshift.
115 This galaxy has I-K>2.6. Again the spectrum shows a well-defined 4000{AA} 
115  break and G-band absorption. This indicates that the galaxy is old and 
115  therefore a reasonable identification for the host.
126 There were two I-band identifications for this object one of which has been 
126  identified as a star via spectroscopy. The second candidate is taken as the
126  host galaxy; however, the radio emission does not clearly associate with 
126  this optical/near-IR source by eye. The second candidate is apparently 
126  unresolved in the K-band image; however, it is very close to the star 
126  making it difficult to tell. It is possible that the true ID could be 
126  hidden behind the star.
133 The red colour, I-K>4, of this galaxy, in addition to its position, 
133  coincident with the radio source, makes it the likely host.
135 The candidate found on one component is confirmed in the K band. It has a 
135  colour of I-K>2.5, which is potentially blue.
136 A red source appears in the K band. It is somewhat offset from the main 
136  radio component, but may be reasonable if the radio emission to the south 
136  of the main component is associated with this source.
148 This host galaxy is very close to a brighter nearby object. This means that 
148  the quoted error on the magnitude may be optimistic.
150 There is no clear candidate associated with the source. For now the 
150  brightest galaxy is taken to be the host galaxy.